There has been publicly known an image input apparatus which irradiates a subject with radiation emitted from a radiation generating device for a sickness diagnosis etc., reads the radiographic image formed through the radiation having penetrated the subject, and inputs the image information. Such an image input apparatus includes: a system which condenses photo-stimulated light generated by scanning with a laser beam a photostimulable phosphor plate having the radiographic image information accumulated therein, and performs photoelectric conversion with a photomultiplier (hereinafter may be referred to as PMT) to convert the light into an electric signal (CR type); and a system which has a function to convert X-ray energy irradiated through the subject into an electric signal for re-structuring the X-ray energy as an X-ray penetration image, and configured of a flat-plate-shaped X-ray image flat detection device (X-ray flat panel detector) having a flat face of an area that sufficiently covers a human body part necessary for an image diagnosis (FPD type).
For example, in a CR type image input apparatus, unevenness in laser beam scanning with an optical system and unevenness in radiography sensitivities are generated due to the PMT that performs photoelectric conversion, and therefore, the unevennesses need to be corrected. Conventionally, such correction is performed using correction values created at the time of production of an image input apparatus. In general, correction value creation (calibration) in view of compensating of unevenness due to optical system is performed in accordance with a large cassette size or the like, and, by using the correction value obtained based on the large size, correction is performed over the large size to a smaller size. A plurality of correction values mentioned above may be created, for example, for each of sampling pitch differences, which are one of correction coefficients. However, the other correction coefficients (size, etc.) are collected into one, and therefore, even when reading is performed for a small size at the user's end, calibration has to be performed for the largest size upon creating correction values.